The LGW right lab released schematic-figure, a skill for AI agents that generates publication-quality schematic figures for scientific papers as editable vector graphics (SVG and PDF) with the underlying drawing code. Demos compare Opus 5.5 with and without the skill on five papers, four published after the model's training cutoff. The authors position it as a fast first draft, not a replacement for figure design judgment.
AutoIllustrate is a command-line tool that turns a methodology brief — a paper section, system description, or verbal sketch — into publication-grade academic figures. It uses LLM agents to generate an editable, vectorised .drawio diagram, then an image-generation model to produce a more visually polished render. It targets researchers who need publication-ready, editable figures from methods text.
AutoIllustrate is a command-line tool that turns a methodology brief — a paper section, system description, or verbal sketch — into publication-grade academic figures. It uses LLM agents to generate an editable, vectorised .drawio diagram, then an image-generation model to produce a more visually polished render. It targets researchers who need publication-ready, editable figures from methods text.
MechFig is a web app that generates scientific mechanism-style figures (pathways, process schemes) from text descriptions. It uses AI to help researchers quickly create draft figures for papers and presentations, replacing manual drawing in PowerPoint or Illustrator. The tool offers style presets and customization options.
MechFig is a web app that generates scientific mechanism-style figures (pathways, process schemes) from text descriptions. It uses AI to help researchers quickly create draft figures for papers and presentations, replacing manual drawing in PowerPoint or Illustrator. The tool offers style presets and customization options.
Anthropic's status page logged three elevated-error incidents across Claude models and its platform tooling on consecutive days. Claude Mythos 5.1 and Claude Fable 5.1 saw errors for roughly 30 minutes on October 5, Claude Opus 5.5 errored for about 20 minutes on October 6, and usage-data loading on platform.claude.com and the Admin API failed for about 100 minutes overnight October 6-7. All three were resolved, with Messages API traffic unaffected in the platform incident.
Anthropic's status page logged three elevated-error incidents across Claude models and its platform tooling on consecutive days. Claude Mythos 5.1 and Claude Fable 5.1 saw errors for roughly 30 minutes on October 5, Claude Opus 5.5 errored for about 20 minutes on October 6, and usage-data loading on platform.claude.com and the Admin API failed for about 100 minutes overnight October 6-7. All three were resolved, with Messages API traffic unaffected in the platform incident.
An arXiv paper (Sept 1, 2026) introduces FigTree, a multi-agent system that formulates scientific methodology figure generation as recursive SVG program construction. Instead of raster image generation, it produces editable vector figures iteratively, addressing the difficulty of single-pass, human-satisfactory figures and precise edits.
An arXiv paper (Sept 1, 2026) introduces FigTree, a multi-agent system that formulates scientific methodology figure generation as recursive SVG program construction. Instead of raster image generation, it produces editable vector figures iteratively, addressing the difficulty of single-pass, human-satisfactory figures and precise edits.
Researchers introduced SciDiagramEdit, a benchmark and skill-evolution framework for editing scientific diagrams using natural language instructions. The task is challenging because scientific figures combine heterogeneous elements like schematics, plots, and captions under a strict visual grammar. This work aims to automate the time-consuming revision of figures in research papers.
Researchers introduced SciDiagramEdit, a benchmark and skill-evolution framework for editing scientific diagrams using natural language instructions. The task is challenging because scientific figures combine heterogeneous elements like schematics, plots, and captions under a strict visual grammar. This work aims to automate the time-consuming revision of figures in research papers.